/
Chapter 11 Chapter 11

Chapter 11 - PowerPoint Presentation

jane-oiler
jane-oiler . @jane-oiler
Follow
375 views
Uploaded On 2016-07-04

Chapter 11 - PPT Presentation

Computer Organization and Architecture 2014 Cengage Learning Engineering All Rights Reserved 1 Computer Organization and Architecture Themes and Variations 1 st Edition Clements 2014 Cengage Learning Engineering All Rights Reserved ID: 390683

organization computer rights data computer organization data rights architecture 2014 cengage learning engineering reserved edition clements variations themes disk

Share:

Link:

Embed:

Download Presentation from below link

Download Presentation The PPT/PDF document "Chapter 11" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

Slide1

Chapter 11

Computer Organization and Architecture

© 2014 Cengage Learning Engineering. All Rights Reserved.

1

Computer Organization and Architecture: Themes and Variations, 1

st Edition ClementsSlide2

© 2014 Cengage Learning Engineering. All Rights Reserved.

2

Secondary StorageSecondary storage covers technologies used to store data with access times far longer than those of cache and DRAM.

Secondary storage systems generally have capacities far greater than main stores; for example, 2 TB hard drive compared with 8 GB

DRAMSecondary storage uses magnetic and optical storage technologies, although solid state flash technologies are beginning to replace some magnetic based devices.Slide3

© 2014 Cengage Learning Engineering. All Rights Reserved.

3

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

IBM shipped the first disk drive in 1956 as part of the 305 RAMAC system.

Its

platter was 24 inches in diameter; it had a capacity of 5 MB, was larger than a washing machine and cost thousands of dollars.

By

1983 the first PC disk drive was introduced by Seagate which also stored 5 MB but cost a mere $1,500 and just managed to fit inside a PC.

Today

, hard disk drives have capacities of 4,000,000 MB and some cost less than $50

.Slide4

© 2014 Cengage Learning Engineering. All Rights Reserved.

4

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The

disk drive uses a technology we have understood since the 1940s.

The

magnetic disk is a direct descendent of the phonograph invented by Thomas Edison in 1877.

Edison

originally stored sound along a track on a cylinder covered by tin foil (later wax).

The

Edison phonograph stored sound by physically deforming the side of the groove to store sound vibrations, whereas the magnetic disk stores data by magnetizing the surface of a track.

The

CD/DVD/Blu-ray stores data by changing the optical properties of the surface of a track.Slide5

© 2014 Cengage Learning Engineering. All Rights Reserved.

5

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

A disk drive uses a flat rotating platter is covered with a very thin layer of a material that can be locally magnetized in one of two directions, north-south or south-north.

This

platter rotates under a write head that magnetizes the surface to create a circular track of 1s and 0s.

When

the data is retrieved, a read head that’s normally co-located with the write head detects the magnetization of the surface and uses it to reconstruct the recorded data.

In

practice, the construction and operation of real disk drives is immensely complex, because the size of the magnetized regions is very small and the disk rotates at a high speed. Slide6

© 2014 Cengage Learning Engineering. All Rights Reserved.

6

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure

11.1

plots areal density against the year for disk drives and illustrates the

increase

in recording density in the 45 years from

1956.

Areal

density

specifies the

density

of bits and is often measured in bits per square inch.

Areal

density increased from approximately 2 x 10

-3

Mbits

/in

2

to 4 x 10

4

Mbits

/in

2

, an increase of the order of 10

7

in

15

years.

By

2010 Toshiba was shipping disks with an areal density of 540

Gbits

/in

2

(i.e., 54 x 10

4

Mbits

/in

2

) and only one year later

announced

a 2.5 inch hard drive for use in laptops with a density of 744 

Gbits

/in

2

.Slide7

© 2014 Cengage Learning Engineering. All Rights Reserved.

7

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Disk capacity has gone hand in hand with the development of Microsoft’s operating systems.

Figure

11.2 demonstrates how both disk capacities and operating systems have grown. Slide8

© 2014 Cengage Learning Engineering. All Rights Reserved.

8

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Magnetism and Data Storage Magnetic storage technology is rather

quaint; it belongs to the early days of computer technology and relies on electromechanical mechanisms with moving parts. Magnetic recording techniques have been used for a long time; for example, the wire sound recorder recorded speech on a reel of steel wire before being replaced by the tape recorder after WW2.

However, magnetic storage technology stubbornly refuses to go away in spite of its inherent limitations. The

magnetic properties of matter are probably the most obvious means of storing data because magnetism is an excellent binary recording medium: magnetic particles can be magnetized North-South or South-North.

When

certain substances are magnetized, they remain magnetized until they are magnetized in the opposite sense, which makes magnetic storage mechanisms inherently non-volatile

.Slide9

© 2014 Cengage Learning Engineering. All Rights Reserved.

9

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The origin of magnetism lies in the atomic structure of matter; in particular, the behavior of electrons in atoms.

An electron has two motions, its orbit round the nucleus and its spin. The principal cause of magnetization is the spin of electrons.

Electrons have two quantized spin values that are called spin up and spin down.In a ferromagnetic material the spins of individual atoms couple; that is, there is an interaction between neighboring atoms. When an external magnetic field is applied, ferromagnetic atoms tend to align with the field.

When

the external magnetic field is removed, a ferromagnetic material can retain some of the magnetization in the direction of the applied field

.

The quantum interactions between electrons in a ferromagnetic material have a range that extends beyond the individual atoms.

This

interaction causes the magnetic moments of atoms within a region called a

domain

to align in parallel. Domains vary in size from 30 nm to 150 mm

.Slide10

© 2014 Cengage Learning Engineering. All Rights Reserved.

10

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

In a bulk ferromagnetic material, the individual domains are aligned at random as

Figure

11.3(a) illustrates and there’s no overall magnetization. Figure 11.3(b) demonstrates the effect of applying an external field. Domains

that are magnetized in the same direction as the external field remain magnetized in that direction.

Domains

that are magnetized in other directions rotate the direction of their magnetization in the direction of the external field.Slide11

© 2014 Cengage Learning Engineering. All Rights Reserved.

11

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The

internal field

is the sum of the external field and the field due to the domains and more and more domains become oriented in the direction of the external field.

Suddenly

, the number of domains rotating in the direction of the external field increases

as

the internal field builds up like an avalanche.

Soon

, all domains are magnetized in the same direction Figure 11.3(c) and bulk material is said to be magnetized. If the external field is removed, the material remains in the magnetized state because the fields from the domains are sufficient to keep the domains from realigning themselves. Slide12

© 2014 Cengage Learning Engineering. All Rights Reserved.

12

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Plotting the

internal field against the external field for a ferromagnetic material,

the hysteresis curve of Figure 11.4.

The horizontal axis, H, represents the external field. In the absence of an external

field,

the

internal field

is either +

B

m

or –

B

m

; that is, the material is magnetized in one of two

states.Slide13

© 2014 Cengage Learning Engineering. All Rights Reserved.

13

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The Read/write Head

 

Figure 11.5 describes the structure of a read/write head used for writing and reading data on magnetic recording media. Very

early recording heads consisted of

toroids

of magnetically soft metallic

ferromagnets

with a few turns of wire around them.Slide14

© 2014 Cengage Learning Engineering. All Rights Reserved.

14

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

High-frequency magnetic fields induce eddy currents

in the write heads that reduce the head’s efficiency. Second-generation heads used non-conducting ceramic ferromagnetic materials called ferrites.

Constructing read/write heads is difficult because of the complex interacting requirements of the recording system. The air gap must be narrow. If the gap is large, the area of magnetization is also large and the number of bits that can be stored is reduced.

It’s also necessary to build the heads with a ferromagnetic material that has a very high

saturation.

A strong field is required

to

magnetize the particles on the recording medium.

The

head material must also have a low remnant magnetization (i.e., the residual field after the write current has been turned off). If the remnant magnetization is too high, previously magnetized bits can be disturbed by the remnant field. Slide15

© 2014 Cengage Learning Engineering. All Rights Reserved.

15

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

If the same head is used for reading as well as writing, it must have a high value of

permeability (a material’s permeability is a measure of its ability to conduct magnetic flux, the higher the permeability the easier it is to magnetize a material). Read heads should also have a low saturation

magnetostriction. Magnetostriction describes a phenomenon whereby a change in magnetic field changes the material’s physical dimensions, and vice versa.

If

a read head suffers any form of physical shock,

magnetostriction

generates a spurious field and hence a spurious current in the coil.

On

top of all these magnetic properties, the head must be physically robust and resistant to both wear and corrosion.Slide16

© 2014 Cengage Learning Engineering. All Rights Reserved.

16

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure

11.6 describes the recording process. A coil of wire is wound round a ring of metal. When a current is passed through the coil, a magnetic field is created in the coil and this, in turn, induces a field in the ring.

The ring contains a tiny air gap and the field has to flow across the gap. In practice, it leaks out into the surrounding world. If the gap is close to a ferromagnetic material, this external field can magnetize it. Slide17

© 2014 Cengage Learning Engineering. All Rights Reserved.

17

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.7 illustrates the effect of switching the current in the write-head coil and the corresponding magnetization of the surface passing under the write-head.Slide18

© 2014 Cengage Learning Engineering. All Rights Reserved.

18

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Limits to Magnetic Recording Density 

There are physical limitations on the ultimate areal density of magnetic recording mechanisms; that is, there is a finite limit to the maximum number of bits that can be stored in a square inch. As the anticipated

theoretical limit is reached, physicists and engineers seem to find ways of extending this theoretical maximum (echoes of Moore’s law). In the 1990s scientists believed that the superparamagnetic effect imposed a limit on magnetic recording density. The minimum amount of magnetic material that can be used to store information is the magnetic

grain

from

which bulk magnetic materials such as the surface of a disk are constructed.

When

grains reach sizes of the order of 10 nm or so, thermal effects can cause these grains to spontaneously demagnetize at room temperature.

This

corresponds to a maximum areal density of about 6 Gb/cm

2

or 0.93 Gb/in

2

Slide19

© 2014 Cengage Learning Engineering. All Rights Reserved.

19

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Limits to Magnetic Recording Density

 

Figure 11.8 illustrates surface recording with horizontal magnetization. This figure represents the worst case with the smallest possible regions of magnetization, magnetized alternately NS and SN.

An

alternative to horizontal magnetization is the vertical or perpendicular magnetization of Figure Slide20

© 2014 Cengage Learning Engineering. All Rights Reserved.

20

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

An alternative to horizontal magnetization is the vertical or perpendicular magnetization of Figure 11.9 in which the magnetic domains are magnetized at right angles to the surface of the recording medium.

Perpendicular

recording reduces the demagnetizing influences of adjacent bits because they are oriented so they do not oppose each other; they form part of a closed magnetic field

.

A

different write head

structure is

required for writing. A so-called monopole head is required to write magnetic transitions vertically within the media

.Slide21

© 2014 Cengage Learning Engineering. All Rights Reserved.

21

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The proximity of the tiny magnetic particles to each other tends to demagnetize adjacent particles.

Figure

11.10 illustrates a means of reducing the size of vertical particles without demagnetization developed by Fujitsu.

This

technology can provide an eight-fold increase in areal density over conventional techniques and permit densities in the region of 50 Gb/cm

2

. By about 2000 esoteric magnetic media with areal densities approaching 700 Gb/cm

2

(100 Gb/in

2

) were being proposed.

By

2011 disks with area densities of over 700 Gb/in

2

were being

sold

.Slide22

© 2014 Cengage Learning Engineering. All Rights Reserved.

22

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

One approach to delaying the inevitable effects of

superparamagnetism

is to modify the media’s magnetic properties by increasing the energy barrier required to reverse the state of a grain. The arrangement in Figure 11.10 uses a layer of magnetically soft material beneath the magnetic recording surface.Slide23

© 2014 Cengage Learning Engineering. All Rights Reserved.

23

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Principles of Data Recording on Disk

 

The next step is to look at how data is stored on a disk and how the recording and playback process works. Figure 11.11 shows how data is arranged on the surface of a disk.

The

read/write head

steps

in towards the center or out towards the periphery. As the disk rotates, the head describes a circle, called a

track

.

A

track is too large a unit of data to be practical, so the track is divided into individual

sectors

. A sector is the smallest unit of data that can be read from or written to the disk. Slide24

© 2014 Cengage Learning Engineering. All Rights Reserved.

24

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The

structure of data on a disk has important implications for the performance of disk drives.

Consider the granularity of data; small sectors are inefficient because a large file would take up many sectors, each of which has an overhead.

Large sectors

are inefficient if you wish to store small units of data; for example, if sectors were 8 KB and you were using a lot of 3 KB files, each sector would waste 5 KB. Typical disk drive sectors are 512 bytes.Slide25

© 2014 Cengage Learning Engineering. All Rights Reserved.

25

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Platter Size

 

The standard hard disk in a PC is called a 3.5 inch drive.

You

would expect the platter to be less than 3.5 inches.

It

isn’t.

The

diameter of a platter is 3.75 inches.

And

, yes, a 3.5 inch drive isn’t really 3.5 inches wide; it’s four inches wide.

The

term 3.5 inches refers to its

form factor

and not its physical width.Slide26

© 2014 Cengage Learning Engineering. All Rights Reserved.

26

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

To reduce

the physical size of a disk drive and increase its data capacity, disk manufacturers

co-locate several platters on the same spindle and the read/write heads that read each surface are connected to the same actuator so that all heads step in or out together.

Early

disk drives didn’t use the top- and bottom-most surfaces to store data and the arrangement of Figure 11.12 has four surfaces.

However

, modern disk drives use all surfaces

.Slide27

© 2014 Cengage Learning Engineering. All Rights Reserved.

27

Computer Organization and Architecture: Themes and Variations, 1st Edition ClementsSlide28

© 2014 Cengage Learning Engineering. All Rights Reserved.

28

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.14 illustrates the way in which the surface-head gap has been reduced between 1993 and 2004.

The

improvement is somewhat under an order of magnitude with contemporary heads tracking a disk surface at a height of about 10 nm (i.e., 10-8

m), an unimaginably small gap.

 Slide29

© 2014 Cengage Learning Engineering. All Rights Reserved.

29

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Should the head fail to follow the surface of a disk and hit the surface at about 50 mph, it damages the magnetic coating and destroys data. Such an event is called a

head crash

, and is the origin of the term now used to indicate any sudden and catastrophic computer system failure.

Figure

11.15 gives

ad illustration

of just how small the read/write head to surface gap is. On the same scale we have the gap together with a human hair, a smoke particle, and a fingerprint. Slide30

© 2014 Cengage Learning Engineering. All Rights Reserved.

30

Computer Organization and Architecture: Themes and Variations, 1st Edition ClementsSlide31

© 2014 Cengage Learning Engineering. All Rights Reserved.

31

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Some platters are made of

glass because it is more thermally stable, smoother, and harder than aluminum. Figure 11.17 is a photomicrograph from IBM showing the relative surface smoothness of both aluminum and glass surfaces.

Glass

is more rigid than aluminum for the same weight of material. Improved rigidity reduces noise and vibration at high speeds. The rigidity of glass allows platters to be made thinner and lighter which reduces the load on spindle motors. Slide32

© 2014 Cengage Learning Engineering. All Rights Reserved.

32

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Modern platters contain five or more layers (Figure 11.18).

The

uppermost layer is a lubricating layer that enhances the durability of the head-disk interface.

Below

the lubricating layer lies a thin protective carbon-based overcoat. The lubricating layer is about 1 nm thick and the overcoat 15 nm thick.

The

recording surface consists of two layers; the recording layer (often a compound of cobalt and chromium) and a chromium

underlayer

.

A glass

substrate provides the surface that holds these other five layers.Slide33

© 2014 Cengage Learning Engineering. All Rights Reserved.

33

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The GMR Head – a Giant Step in Read Head Technology 

The conventional read head has an important limitation, inductance. In order to detect the changing magnetic flux, the coil round the head requires a lot of turns to pick up sufficient signal. Increasing

the number of turns raises the coil’s inductance. Inductance is a property of circuits that resists the rate at which a current changes. Increasing the inductance of a read head reduces the rate at which it can read changes in the magnetic flux at the disk’s surface

Fortunately

, another magnetic property of matter was discovered that made it possible to do away with the inductive read head.

Moreover

, removing the read head means that head can be optimized for writing.Slide34

© 2014 Cengage Learning Engineering. All Rights Reserved.

34

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

A

magnetic field causes a tiny change in the electrical resistance of certain materials, a property called the

magnetoresistive effect, MR. Detecting

changes in magnetic flux from a disk using the MR effect has advantages over

inductive

read-heads because the inductance of an MR head is lower and it’s possible to read data more rapidly. Slide35

© 2014 Cengage Learning Engineering. All Rights Reserved.

35

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The GMR Head – a Giant Step in Read Head Technology

 

In the late 1980s, researchers discovered that some materials exhibited massive changes of up to 50% in their resistivity in the presence of a magnetic field. This property was called the giant magnetoresistive

(GMR) effect and is found in materials consisting of alternating very thin layers of metallic elements. Slide36

© 2014 Cengage Learning Engineering. All Rights Reserved.

36

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

In a read/write head

assembly

the read element consists of a GMR sensor between two magnetic shields. These magnetic shields reduce unwanted magnetic fields from the disk so that the head detects only the magnetic field from the recorded data bit under the head. In a merged head a second magnetic shield also functions as one pole of the inductive write head. The advantage of separate read and write elements is that both elements can be individually optimized. Slide37

© 2014 Cengage Learning Engineering. All Rights Reserved.

37

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Pixie Dust At the beginning of 2001 IBM announced a breakthrough in disk technology that could increase areal densities by a factor of four. IBM used a sandwich with three layers to store data.

The top layer is a ferromagnetic material that stores the data. The lower layer is an

antiferromagnetic layer. Antiferromagnetism occurs when atoms align themselves antiparallel to a magnetic field, the opposite of ferromagnetism. However, antiferromagnetics is a very weak effect.

Between

these two layers sits

pixie dust

, a three atom thick layer of the element ruthenium

.

Ruthenium

is a rare metal belonging to the same group as platinum and only about twelve tons are produced

annually.Slide38

© 2014 Cengage Learning Engineering. All Rights Reserved.

38

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

This

sandwich is called an

antiferromagnetically-coupled (AFC) media and is capable of areal densities of up to about 100 Gb per inch2. IBM claims that AFC media avoids the high-density data decay.

The

ultra-thin ruthenium layer forces the adjacent layers to orient themselves magnetically in opposite directions.

The

opposing magnetic orientations make the entire multilayer structure appear much thinner than it actually is. Thus, small, high-density bits can be written easily on AFC media, but they will retain their magnetization due to the media's overall thickness. Slide39

© 2014 Cengage Learning Engineering. All Rights Reserved.

39

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.21 illustrates the principal of the optically assisted write head. A tiny laser beam performs the surface heating.

An

optical fiber delivers the laser beam to the disk’s surface via an optical fiber and a lens assembly.

Pulsing

the laser beam allows you to selectively heat the surface. A current is passed through the coil above the surface of the disk to magnetize the heated spot in one sense or another.Slide40

© 2014 Cengage Learning Engineering. All Rights Reserved.

40

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

When a read head passes over a magnetized surface, the

changing

magnetic flux induces a current in the coil and a voltage across the coil’s terminals. The voltage is proportional to the rate of change

of the magnetic

flux;

that is, you can detect only a

change

in flux density

.

Figure 11.22 illustrates the write current in the write

and

the resulting magnetization of the recording surface. Below this

is a

trace of the voltage induced in the coil when the recorded surface passes under the head. Slide41

© 2014 Cengage Learning Engineering. All Rights Reserved.

41

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

You can’t store a long string of ones and zeros on the surface reliably because only changes in flux lever create a signal in the head.

If you record 00000 or 11111 both sequences would produce the same output, nothing. Suppose, for example, you stored the string 00

0111111111110000, the read head would detect only two flux transitions, the initial 0 to 1 and the final 1 to 0.

A GMR head can detect absolute magnetization because even a constant field creates detectable low or high resistance in the

magnetoresistive

element. However, there is no delineation between the ones and zeros in a long string.Slide42

© 2014 Cengage Learning Engineering. All Rights Reserved.

42

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Digital data recording mechanisms (both magnetic and optical) encode data prior to recording in order to avoid situations in which the recorded information is difficult to read back.

In particular, they avoid longs runs or constant magnetization; that is, they ensure that the recoded flux changes state regularly. This

restriction is required to extract a data clock from the disk. Most recorded codes are known as self-clocking because the bit pattern recorded on the magnetic surface contains sufficient information to allow hardware to recreate or regenerate

a clock wave that can be used to sample the incoming data.Slide43

© 2014 Cengage Learning Engineering. All Rights Reserved.

43

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The design of codes for recording data is a fine art because of the many conflicting requirements.

You want to increase the efficiency of the code by reducing the number of flux reversals required to record each bit (the best you can do is one transition per bit). You

want to make the recorded symbols for 1 and 0 as unlike each other as possible in order to make it easy to tell the difference between 1s and 0s in the presence of noise and other extraneous signals.You want to ensure that there is no significant gap between flux transitions to make the code self-clocking; you want to avoid patterns that contain low-frequency components because the analog circuits that process data from the read head do not handle low frequencies well.Slide44

© 2014 Cengage Learning Engineering. All Rights Reserved.

44

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Any recording code is a compromise. Figure 11.23 illustrates an encoding method once used by floppy disk drives called

modified frequency modulation

, MFM. Floppy disk drives operate on the same principles as the hard disk except that the head is in contact with the recording surface and the speed of rotation is very much slower.Slide45

© 2014 Cengage Learning Engineering. All Rights Reserved.

45

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.23 shows the sequence of bits to be recorded; that is, 010100111. A clock pulse marks the boundary between each of these bits.

The

first step in the encoding process (line labeled Data pulses

)

is to generate a pulse whenever the data bit to be stored is a 1.

If

these pulses were used to store data directly, a problem would occur whenever the input stream contained two or more consecutive 0s, because there would be no recorded data. Slide46

© 2014 Cengage Learning Engineering. All Rights Reserved.

46

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

MFM solves the problem of a lack of signal when there is a stream of 0s by recording a 1 at the cell boundary between two consecutive 0s.

This

rule ensures that a string of 0s still creates flux transitions, and yet the inserted pulse is not interpreted as a 1 because it falls between cell boundaries rather than in the middle of a cell.

The

bottom line of Figure 11.23 is the current in the write head that changes direction on each pulse.Slide47

© 2014 Cengage Learning Engineering. All Rights Reserved.

47

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.24 illustrates the structure of a track. A sector should be as large as possible for the purposes of storage efficiently.

Since

each sector contains housekeeping information, small sectors are inefficient because they waste disk space. A sector should be as small as possible because of storage efficiency. Since a sector is the smallest unit of data that can be written, the unit of granularity of a file is the sector. Slide48

© 2014 Cengage Learning Engineering. All Rights Reserved.

48

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.25 shows the structure of a track that is written to the disk when the disk is first

formatted

(this is a floppy disk sector structure which is easy to understand). Until this structure is laid down, the disk cannot be used to record data.

The

overhead needed to store data has no equivalent in semiconductor memories.Slide49

© 2014 Cengage Learning Engineering. All Rights Reserved.

49

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Zoning

 

The circumference of a track is

p

·

d

, where

d

is the diameter of the track. It there are

n

sectors per track, the length of a sector is

p

·

d

/

n

. The approximate size of a bit is given by

p

·

d

/(

m

·

n

)

, where

m

is the number of bits per sector

.

Because

the value of

d

varies radically between the inner and outer tracks, the width of a bit varies correspondingly. If the size of a bit is sufficiently large to be detected on the innermost track, it is too large on the outermost track and the storage efficiency is compromised

.

Disks

deal with the problem of different track lengths by

zoning

, whereby adjacent tracks are grouped into zones and each zone has a different number of sectors. Some of today’s disks divide the surface into 30 or more zones.

Zoning

affects the rate at which data is read from the disk. With fewer sectors along the innermost track, the data transfer rate may be 60% less than when reading a sector at the outermost edge.

 Slide50

© 2014 Cengage Learning Engineering. All Rights Reserved.

50

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.26 demonstrates how a file is

read.

A file is composed of a sequence of sectors. The sectors themselves may be arranged as a linked list or a directory may define the sequence of sectors belonging to a file.

When

files are first created they are allocated sequential sectors. After a period of file creation and deletion, the free sectors on a disk become highly scattered and the resulting files heavily fragmented.

Operating

systems can either automatically or manually defragment files by periodically reorganizing their structure to minimize the seek time.Slide51

© 2014 Cengage Learning Engineering. All Rights Reserved.

51

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The access time of a disk is composed of two major components: the time taken to access a given track (the

seek time) and the time to access a given sector once its track has been reached (the latency).

The latency is easy to calculate. Assuming that the head has stepped to a given track, the minimum latency is zero (the sector is just arriving under the head). The worst-case latency is the period of one revolution (the head has just missed the sector and has to wait for it to go round).

On

average, the latency is 1/2t

rev

, where

t

rev

is the time for a single revolution of the platter. If a disk rotates at 7,200 rpm, its latency is given by:

 

½ x 1/( 7,200

60) = 0.00417s = 4.17

ms.Slide52

© 2014 Cengage Learning Engineering. All Rights Reserved.

52

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

What is the average seek time? Suppose the disk has

N

sectors and it takes tstep seconds to step from track-to-track.

If

the head is parked at the edge of the disk after each seek, the average number of tracks to step over when seeking a given track would be

N

/2, and the average seek time is given by ½ x

N

x

t

step

.

Figure

11.27(a) illustrates this situation

.Slide53

© 2014 Cengage Learning Engineering. All Rights Reserved.

53

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

However

, suppose that the head were automatically moved to the center of the tracks (track number

N/2) after each access - Figure 11.27(b).

When

a new seek operation is issued, the head can move either left or right. In this case the average number of tracks to step is

N

/4, and the average seek time becomes ¼ x

N

x

t

step

.Slide54

© 2014 Cengage Learning Engineering. All Rights Reserved.

54

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.28 demonstrates the movement of an arm with a rotary actuator where it spends part of its time accelerating up to speed, part of its time moving at approximately common speed across the disk’s surface, and part of its time decelerating as it approaches its target destination.Slide55

© 2014 Cengage Learning Engineering. All Rights Reserved.

55

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Accessing Sectors

 

Suppose an operating system makes a series of requests to a disk drive for sectors on tracks 50, 150, 32, 16, 125, 8, 130, 50, 60, 200. Figure 11.29 provides a head movement graph of the time plotted against the position of the head if the requested tracks are accessed in the order in which they were received; that is

first-come, first-served

(FCFS).Slide56

© 2014 Cengage Learning Engineering. All Rights Reserved.

56

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Operating systems often provide support for disk drives. For example, Figure 11.30 illustrates the same situation as in

Figure

11.29 except that the tracks are buffered and the surface of the disk is first swept in one direction and then back in the other direction. Slide57

© 2014 Cengage Learning Engineering. All Rights Reserved.

57

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

 Algorithms for scanning

FIFO First-in-first out (also called first come first served, FCFS). This algorithms processes requests in the order in which they are received. It is fair to all processes. Sequence = 10, 19, 3, 14 ,12, 9. Average seek length = 8.2

.SSTF Shortest seek time first. The next seek is the one that is closest to the current head position. This algorithm selects the next request as the one requiring the least movement of the head. Because newer requests may be serviced before older requests, it is not fair. Sequence = 10, 12, 14, 19, 9, 3. Average seek time = 5.0.

SCAN The approach implements the so-called

elevator algorithm

by taking the closest request in the direction of travel. It satisfies all outstanding requests in the current direction of head motion before reversing direction. Sequence = 10, 14, 19, 12, 9, 3. Average seek time = 5.0.Slide58

© 2014 Cengage Learning Engineering. All Rights Reserved.

58

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

LOOK This is a modest variation on SCAN where the software looks ahead and changes direction when there are no requests beyond the current sector

.C_SCAN The circular scan algorithm moves the head in one direction, sweeping across the disk. However, the head then moves back to the other edge of the disk and starts again. It is a unidirectional version of SCAN. Sequence = 10, 14, 19, 3, 9, 12. Average seek time = 6.8

.FSCAN This algorithm is intended to deal with arm stickiness. Two request queues are required. Initially, at the start of a scan all requests are in one queue and the other queues is empty. Once a scan is in progress, all new requests are put in the other queue. This mechanism defers all new requests until the existing ones have been serviced. It is, of course, a fair access mechanism.Slide59

© 2014 Cengage Learning Engineering. All Rights Reserved.

59

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

SMART Technology 

Hard disks are complex electromechanical devices operating at the frontiers of technology. Electromechanical systems with moving parts are far more unreliable than their semiconductor counterparts.

The major disk manufacturers developed a technology called SMART that monitors the performance of hard disk drives and can predict the probability of failure and therefore provide the user with an advance warning of possible failure. The acronym SMART stands for

self-monitoring, analysis and reporting

technologySlide60

© 2014 Cengage Learning Engineering. All Rights Reserved.

60

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Some of a disk drive’s parameters that can be used to indicate possible failure are 

head flying heightdata throughput

spinup timere-allocated sector countseek error rate

seek time performance

spin try recount

drive recalibration recount

 Slide61

© 2014 Cengage Learning Engineering. All Rights Reserved.

61

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Effect of Temperature on Disk Reliability

 

The reliability of both the electronics and the mechanics such as the spindle motor and actuator bearings degrades as temperature rises. Operating a disk drive at extreme temperatures for long periods dramatically reduces its life expectancy.

Figure

11.31 shows the relationship between the temperature and reliability of a hard disk drive.

 Slide62

© 2014 Cengage Learning Engineering. All Rights Reserved.

62

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Shake, Rattle and Roll

 

Mechanical

systems are sensitive to

disturbances

such as vibration.

It

is not easy to keep

heads

centered over tracks

spaced

at a pitch of 50 nm (2,000 times narrower than a human hair at 100 microns) on

rotating disks.

Disk drives use feedback

to

center the head over a track; that is, an error signal due to

mispositioning

of the head is used to move the

to

reduce the error.

When

a drive suffers from external vibration, the head moves off position and generates a bigger error signal which moves the head back. Unfortunately, this is a

post hoc

solution; the correction happens

after

the disturbance.

An

Hitachi white paper (Rotational Vibration Safeguard) describes a new technique used to minimize the effects of vibration.

Two

vibration sensors

are

located on the disk drive’s

board

to detect movement. The signals from these sensors

move

the head actuator

back

into position

before

the head has wondered off track. In actual tests where a disk was shaken to reduce performance to about 30%, the application of the rotation vibration safeguard mechanism increased performance to 90%.

 Slide63

© 2014 Cengage Learning Engineering. All Rights Reserved.

63

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Secure Memory and RAID Systems 

in 1987, Patterson, Gibson, and Katz at UC Berkeley published, A Case for redundant array of inexpensive disks, RAID, proposing a means of exploiting the inexpensive disk systems found in PCs.

The expression array of inexpensive disks implies a regular structure built around commodity, off-the-shelf disk drives, and redundant implies a degree of fault tolerance; that is, the failure of a single drive should not bring down the entire system.

The

RAID concept rapidly moved out of the

laboratory

and by the mid-90s, RAID systems were being advertised in personal computer magazines.

Today

, most PCs have motherboards with RAID support.

Patterson

et al proposed several ways of organizing clusters of drives at Level 0 RAID, Level 1 RAID, etc. The various RAID levels provide different functionalities; some emphasize speed and some

reliabilitySlide64

© 2014 Cengage Learning Engineering. All Rights Reserved.

64

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Disk drives are inherently serial storage devices. In a RAID system multiple disk drives are operated in parallel and the bits of a single file are divided between the individual drives.

RAID arrays can be used to improve the performance of disk systems or to improve their reliability

. By replicating data across several disks, the failure of a single disk can be made invisible to the system. RAID systems are important in applications where data security is vital (e.g., banking

).

The reliability of

n

disk drives is about 1/

n

that of a single because there are

n

of them to fail.

If

data is distributed between the disks of an array, the failure of one disk doesn’t cause the system to crash. This statement is not true of Level 0 RAID, which is a special case. Slide65

© 2014 Cengage Learning Engineering. All Rights Reserved.

65

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

To understand RAID, we need to remember two concepts.

First, the data recorded on disks is stored in sectors. Second, each sector has a

frame check sequence that can detect one or more errors in a sector. This ability to detect errors means that a RAID array can rapidly respond to the failure of an individual drive unit. The

key concept in RAID technology is

striping

. Disk space is divided into units called

stripes

that may be as small as a single sector or as large as several megabytes.

These

stripes are spread out or

interleaved

across several disks in parallel.

The

way in which data is allocated to the stripes, and the stripes allocated to the individual drives determines the

level

of the RAID system.Slide66

© 2014 Cengage Learning Engineering. All Rights Reserved.

66

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

RAID

level 0

is the only level not using redundancy to provide additional security. A Level 0 RAID with n drives divides data into n stripes that are applied in parallel to the

n

drives.

Figure

11.32 demonstrates Level 0 striping with four drives.

The

size of each disk drive should be the same; otherwise the effective size of a RAID array defaults to the size of the smallest drive in the array.Slide67

© 2014 Cengage Learning Engineering. All Rights Reserved.

67

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The advantage of a RAID 0 array is its high throughput. For n drives, the capacity is

n times that of a single drive and the speed is higher because read operations can take place in parallel. The RAID controller may be implemented in either hardware or software. However, as we said earlier, it is common for RAID controllers to be built into PC motherboards leaving the user with little to do other than to plug in the disk drives and configure the BIOS.

The capacity of a RAID 0 array with n drives is simply n times the capacity of one drive; that is, no capacity is lost due to redundancy and the storage efficiency is 100%. However, there is no fault tolerance and the loss of any drive in the array renders all the data invalid. Read and write performance is excellent

.

Because the loss of one disk brings down the entire system, a RAID 0 array makes sense only if the data is frequently backed up and provision is made for failure between backups. In that case, RAID 0 is both efficient in terms of its use of disk space and fast because of its inherent parallel read and write mechanism

.Slide68

© 2014 Cengage Learning Engineering. All Rights Reserved.

68

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.33 illustrates a level 1 RAID

mirroring

array replicates copies of stripes on multiple drives (here we are using only two drives in parallel). Data

security is increased because you can remove one of the drives without losing data.

Level 1

RAID system increase data

security and improve

access time. Suppose a given stripe is accessed

.Slide69

© 2014 Cengage Learning Engineering. All Rights Reserved.

69

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The write time is the longer of the two parallel writes. Fortunately, most accesses are reads rather than writes.

Moreover

, it is possible to cache writes and allow the disk to do the writing when it is free.

A

level 1 RAID array is expensive because it duplicates data, but it can provide cost-effective security in an age where a large-capacity high-speed disk is of the order of $100Slide70

© 2014 Cengage Learning Engineering. All Rights Reserved.

70

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The efficiency of a two-disk RAID 1 system is 50% because data is simply duplicated.

Duplication

provides excellent fault tolerance. If a drive fails, they system can continue working normally.

All

you have to do is to remove the failed drive, install a new one, and then rebuild the lost data. Slide71

© 2014 Cengage Learning Engineering. All Rights Reserved.

71

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

A hybrid

system called

RAID 0+1 or RAID 0/1 combines features of levels 0 and 1, by providing both fast data access and protection against drive failure

.

Figure

11.34 illustrates a system with two sets of three drives. A stripe is written across drives 1, 2, and 3 to provide a level 0 RAID service. However, because drives 1, 2, and 3 are mirrored as 4, 5, and 6, the arrangement provides the security of a level 1 system.

This

arrangement is the most costly form of RAID and upgrades require disks to be added in pairs.Slide72

© 2014 Cengage Learning Engineering. All Rights Reserved.

72

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

RAID Level 2 and Level 3

 

RAID levels 2 to 6 all distribute data in the form of stripes like level 0, but they do not fully replicate the data like level 1. In other words, levels 2 to 6 fall between the extremes of level 0 (no redundancy) and level 1 (redundancy by replication).

RAID levels 2 and 3 employ multiple synchronized disk drives; that is, the spindles are synchronized so that sector

i

passes under the read/write head of each of the disks in the array at the same time. Slide73

© 2014 Cengage Learning Engineering. All Rights Reserved.

73

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Level

2 and 3 arrays provide true parallel access in the sense that, typically, a byte is written to each of the disks in parallel.

The difference between levels 2 and 3 is that level 2 uses a Hamming code to provide error detection and correction, whereas level 3 provides only a simple parity-bit error detecting code.

The

parity check data in a level 3 RAID is stored on one disk, whereas the Hamming

code of a level 2 RAID may be spread over more than one drive. Figure 11.35 illustrates the concept of RAID level 3 which is also

called bit-interleaved parity

.Slide74

© 2014 Cengage Learning Engineering. All Rights Reserved.

74

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

A single-bit parity code can’t normally be used to correct an error. But it can in a RAID level 3 array. Suppose a disk drive fails.

The stripes recorded on the failed disk are, therefore, missing. However, the stripes on the parity disk can be used to reconstruct the missing data.

Table 11.3 illustrates a level 3 array with four data disks and a parity disk. If disk 3 fails, we have the situation of Table 11.4.Because we know the parity bit across each row, we can recalculate the missing data. For example, in line 1, the bits are 0,1,?,0,1. Since the parity bit is odd, there must be an odd number of 1’s in the data bits. Therefore, the missing bit must be 0.

bit 1

bit 2

bit 3

bit 4

P

0

1

0

0

1

1

1

0

0

0

0

1

1

1

1

1

0

1

0

0

bit 1

bit 2

bit 3

bit 4

P

0

1

?

0

1

1

1

?

0

0

0

1

?

1

1

1

0

?

0

0Slide75

© 2014 Cengage Learning Engineering. All Rights Reserved.

75

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

RAID Level 4 and Level 5

 

RAID levels 4 and 5 are similar to levels 2 and 3. However, in these cases the individual disks are not synchronized and operate independently of each other. The

stripes are much larger than levels 2 and 3. In level 4,

block interleaved parity

, the parity stripes are stored on a single disk, whereas in level 5 the parity stripes are interleaved and stored on all disks in the array

.

You can update the parity information in RAID 5 systems more efficiently by changing the corresponding parity bits only when the data changes

.Slide76

© 2014 Cengage Learning Engineering. All Rights Reserved.

76

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

RAID

level 5, Figure 11.36, is a popular configuration, providing striping as well as parity for error recovery.

The parity block is distributed among the drives of array, which gives a more balanced access load across the drives. A minimum of three drives is required for a level 5 RAID array. Slide77

© 2014 Cengage Learning Engineering. All Rights Reserved.

77

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Dealing with Failure in a RAID Array

 

When a hard drive in a RAID array fails, it has to be replaced. RAID controllers can be designed to be hot

swappable. You pull

out the failed drive and insert a replacement without powering down and

rebooting. The

new drive has to be configured for the

array (

rebuilding

).

It’s easy

to rebuild a drive in a RAID 1 system

– you copy

data from a working drive to the new mirror drive. Rebuilding data on a RAID 5 array takes a lot more time because you have to synthesize all the data by reading the appropriate stripes from the other disks and then performing an exclusive OR operation with the data

.

 

RAID 0 The

fastest and most efficient arrangement. No fault-tolerance is provided. Requires a minimum of two drives.

RAID

1 Data

is mirrored (duplicated). This is the best choice for performance-critical, fault-tolerant environments. Requires a minimum of two drives.

RAID 2

This

mode is not used with today’s drives that include embedded ECC mechanisms.

RAID

3 This

mode can be used to speed up data transfer and provide fault-tolerance by including a drive with error correcting information that can be used to

reconstruct

lost data. Because this mode requires synchronized-spindle drives, it is rarely used today. Requires at least three disks.

RAID 4

Little

used.

RAID

5 This

mode combines efficient, fault-tolerant data storage with good performance characteristics. However, performance during drive failure is poor and rebuild time slow due to the time required to construct the parity information. Requires a minimum of three drives

.

  

RAID

0+1 Striped

sets in a mirrored set with four or more disks. Provides fault tolerance and improved performance.

RAID

1+0 Mirrored

sets in a striped set. Like RAID 1+0 but with better performance.

RAID

5+1 A

mirrored striped set. Requires at least three disks.

 

 

 Slide78

© 2014 Cengage Learning Engineering. All Rights Reserved.

78

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Solid-state Disk Drives 

The days of the magnetic drive may be limited by the introduction of solid-state drives that mimic the hard drive electrically. The

solid-state hard drive uses semiconductor flash technology to store data and an electrical interface that makes it physically compatible with hard disk interfaces; that is, you just plug an SSD into a hard disk SATA socket.The SSD has considerable advantages over the electromechanical disks, the most important of which are higher performance, lower power consumption, lower weight, and greater tolerance to shock.

In

2010 SSDs were finding their way into high-end executive laptops (in a 2 ½ inch form factor) and specialist high-end applications in 3 ½ form factors. The limitations of the SSD are twofold; their considerable cost premium over hard drives and their limited storage capacity.Slide79

© 2014 Cengage Learning Engineering. All Rights Reserved.

79

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Solid-state disks are constructed with the type of flash memory technology we discussed in the previous chapter; it’s only in recent years that the cost of flash memory has declined to the point at which large (over 128 GB) memories are economically feasible.

Because solid state disks have no moving parts, they are truly random access devices. There is no rotational latency and no seek time. Consequently, the fragmentation problem associated with hard disks simply goes away. It’s not necessary to periodically defragment an SSD when files are scrambled throughout the memory space.

As early as 2007, a white paper from IDC indicated the potential savings from an SSD-based notebook as: IT labor savings for PC deployment 2.4%

Reliability savings for outsourced repair 1.4%

Reliability savings for repairs 7.5%

Reliability savings for user productivity due to hard drive loss 17.2%

Savings from power enhancements 16.9%

User productivity savings 54.5

%

 Slide80

© 2014 Cengage Learning Engineering. All Rights Reserved.

80

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Magnetic Tape

Magnetic tape provides a means of archiving large quantities of data. In a world where the production of data is estimated to increase by 60% compounded annually, archival storage is vital. In the 1960s and 1970s

the preferred backing medium was magnetic tape because it was relatively cheap and could store large volumes of data.Data is stored on tape as multiple parallel tracks (typically nine; i.e., eight data bits and a parity bit). Magnetic tape recording technology is virtually identical to disk technology except that there is a single track that is one bits wide along a long (typically 2400 feet) flexible magnetic tape.

A

tape drive required large and powerful motors to spin reels of tape rapidly and to stop them equally rapidly.

Magnetic

tape was available on 10.5 inch reels up to the 1980s and data was stored at, typically, 128 characters per inch. Slide81

© 2014 Cengage Learning Engineering. All Rights Reserved.

81

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

All tape systems have a long latency because the tape has to be moved past a read/write head and the desired data may take several seconds or even minutes to locate. Consequently, tape is a purely archival medium

.Tape drives grew smaller and tape cartridges (similar to audio cassettes and VCR tapes) were developed. In order to store more data on the tape the information was stored along diagonal tracks on the tape by using rotating helical read/write heads, the same writing mechanism was used in domestic VCRs.

The QIC (Quarter-inch cartridge) set of standards introduced in 1972 provided a great leap forward over the reel-to-reel machines and supported 1.35 GB tapes with 30 tracks at 51K bits/inch linear density ad a 120 inch/s tape speed. Slide82

© 2014 Cengage Learning Engineering. All Rights Reserved.

82

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

By 2010 the quarter inch cartridge was largely obsolete. Today, linear serpentine

recording is widely used to store data on tape. The term serpentine hints at the zigzag nature of the recording with some tracks recorded left to right and some right to left on the tape.

Consequently, when the tape reaches its end, it does not have to be rewound, but can simply change direction to continue reading or writing data.Slide83

© 2014 Cengage Learning Engineering. All Rights Reserved.

83

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

It’s a popular myth that

tape and cartridge recording technologies are obsolete. However, the death of tape is much exaggerated. In January 2011, HP reported that the worldwide market for their LTO tape drives declined by about 30% between late 2008 and late 2009, but increased by 45% (of the 2008 figure) by the end of 2010.

This amounts to a renaissance in the use of magnetic tape. Moreover, tape has a much lower TCO (total cost of ownership) than disk drives. A large organization such as law enforcement or a medical institution has large quantities of data to store.

The

total cost of ownership includes the equipment, media, floor space,

maintenance,

and

energy

consumption.

The

TCO for data over a 12-year period in a large organization might be fifteen times greater for disk-based storage than for tape-based

storage.Slide84

© 2014 Cengage Learning Engineering. All Rights Reserved.

84

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Towards the end of the 1990s the LTO (linear tape-open) standard was developed.

The word open indicates that the standard is not proprietary like the earlier standards that were owned by IBM or HP.

The first standard, LTO-1, introduced a 100 GB cartridge. By 2010 the LTO-6 standard had been launched and that provided a capacity of 1.5 TB, a data speed of 140 MB/s using 896 tracks at a linear density of 15,142 bits/mm. The LTO standards have been scaled up to 12.8 TB/cartridge in version LTO-8 that has not yet been released. LTO-5 1.5 TB cartridges are not cheap.

In

2011 the cost of a cartridge was of the order of $50 which is compatible with hard-disk storage. Slide85

© 2014 Cengage Learning Engineering. All Rights Reserved.

85

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Manufacturing modern high-density tapes is not easy.

The distance between tracks is so small that tape dimensional stability becomes an issue.

Magnetic tape can change dimension under the influence of heat and humidity. This means that tracks that were aligned with the parallel heads at the time the data was written may not be aligned when the tape is read at a later date. This

has led to the search for highly stable tape substrates; for example, some manufacturers use a polyamide that is a close relative of

Kelvar

. Slide86

© 2014 Cengage Learning Engineering. All Rights Reserved.

86

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Ultrium Technology

 

Ultrium technology is the generic name given to the set of Linear Tape-Open standards developed jointly by Quantum, HP and IBM to provide a common standard in a world where users were often forced to buy a proprietary recording technology that forced them to remain with one supplier

.

The LTO standards have three important features:

They are regularly updated to keep pace with both demand and technological development

They use linear serpentine technology (in contrast with helical recording)

They offer the very high storage capabilities required in today’s world.Slide87

© 2014 Cengage Learning Engineering. All Rights Reserved.

87

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

This slide makes

the point that 90% of network data is never accessed. If data is stored on disks, it takes

up expensive real estate and consuming power. Slide88

© 2014 Cengage Learning Engineering. All Rights Reserved.

88

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Optical Storage Technology 

The optical digital storage medium was an invention waiting to happen and was inevitable. It was just a matter of waiting to see which technology would predominate and which companies would get to the marketplace first.

Here we introduce the three optical recording mechanisms used by digital computers: the CD, DVD, and Blu-ray. We will look at each of these in turn because they all employ the same underlying technology.

The

difference between them is largely one of

scale

; as time has progressed it has become possible to scale down the size of the features on the disk that stores the data.Slide89

© 2014 Cengage Learning Engineering. All Rights Reserved.

89

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Philips developed their LaserDisc

in the early 1970s and launched it in the 1980s. LaserDiscs shared some of the attributes of today’s compact discs CDs, but the LaserDisc used analog technology. The system was read-only and few films ever made it to disc.

During the 1970s Philips and Sony joined forces to develop optical storage; not least because Philips had considerable experience in optical technology and Sony had expertise in encoding and error correction mechanisms.An important achievement of Philips and Sony was the standardization of the

recording

and playback mechanisms.

Two

fundamental parameters are the size of the CD itself and the

sampling frequency

. The disc’s diameter is 12 cm and the sampling frequency was chosen as 44.1 kHz.

Although

this seems a strange number, it’s the sampling rate required for high quality audio and the same frequency is used in television systems. Slide90

© 2014 Cengage Learning Engineering. All Rights Reserved.

90

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Optical Storage TechnologyThe first

practical, low-cost, high-density optical storage mechanism was the CD introduced

in 1981 as a means of storing high-quality sound. The structure of a CD is similar to a magnetic disk because information is stored along a track. A CD’s track is continuous and arranged as a continuous

spiral.

The spiral has about 20,000 turns which corresponds to a length of about 3 miles. The effective track density is 16,000 turns per inch and the theoretical maximum areal density is 1 Mb/mm

2

or 645 Mb per square inch.Slide91

© 2014 Cengage Learning Engineering. All Rights Reserved.

91

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Digital information on a CD is imprinted on the surface of a 1.2 mm thick plastic disk in the form of a pattern that contains indentations, called

pits

, of varying length. The pits are coated with a metallic mirror and a protective layer. The region of the surface that is not indented is called

land

.

Data

is stored from 25mm to 58mm from the center

.

Data is read from the disk's surface by illuminating it with a tiny spot of light and then detecting how much light is reflected back from the surface. A change in reflected light intensity occurs every time the laser spot moves from the pit onto the land and vice versa.Slide92

© 2014 Cengage Learning Engineering. All Rights Reserved.

92

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.40 illustrates the structure of the surface of a CD.

A

pit rises above the surface and it’s what us normal folk call a bump. The term pit refers to an indentation when looking down at the data-carrying layer from above the label side of the disk. The laser beam that reads the data sees a

bump

, rather than a pit.

Figure

11.40

shows four

layers: the disk

is 1.2

 mm thick

transparent

polycarbonate plastic. The pits are coated with a thin layer of aluminum followed by a protective acrylic layer and then the

label.

This label side is much more sensitive to scratches and abrasion than the clear side.Slide93

© 2014 Cengage Learning Engineering. All Rights Reserved.

93

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The areal density of a CD is a product of the number of bits stored along a track and the pitch of the tracks.

The

size of the individual bits is determined by the size of the spot of light projected onto the disk's surface.

Figure

11.41 illustrates the structure and dimensions of the pits and land on a CD’s surface.

A

bump is approximately 0.5 x 10

-6

m and the

track pitch is

1.6 x 10

-6

m. The height of the bump (pit) is 1.25 x 10

-7

m.Slide94

© 2014 Cengage Learning Engineering. All Rights Reserved.

94

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Characteristics of a laser beamSlide95

© 2014 Cengage Learning Engineering. All Rights Reserved.

95

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Abeam

of light illuminates the pits and land along a track and the amount of reflected light is used to read the data.

In order to create and control the smallest possible spot of light, it is necessary that the light beam be both

coherent

and

monochromatic

.

A

light source is

monochromatic

if the light waves all have the same frequency, unlike white light that contains frequencies distributed across the visible spectrum. Slide96

© 2014 Cengage Learning Engineering. All Rights Reserved.

96

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure

11.42 illustrates the energy distribution of a spot of laser light on the surface of a disk. A perfect spot would have hard edges between light and dark.

Because of the wave-like properties of light, the edge of a spot is not sharp; it consists of a series of rings of light.

Figure 11.42

illustrates

the relative dimensions of the spot,

tracks, pits

and land. When light hits a flat region on the

surface,

a large fraction of the light is reflected back. Suppose that the size of the spot is somewhat larger than an individual bump. When light hits a bump on the surface, some light is reflected back from the top of the bump and some from the surface around the bump.Slide97

© 2014 Cengage Learning Engineering. All Rights Reserved.

97

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

In

a CD, the height of the bump above the land is one quarter the wavelength of the laser light.

The light hitting the top of a bump travels x

from

its source

to the surface and then to the observer. The light

hitting

the bottom of the bump

travels a further

l

/4 in each direction.

The

total difference traveled by the light reflected off the land is x +

l

/2. Because the spot doesn’t cover a pit entirely and the path length between the reflected light from the pit and from the land is half the wavelength of the light, the beams tend to cancel. Slide98

© 2014 Cengage Learning Engineering. All Rights Reserved.

98

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.43 demonstrates the how the effects of surface contamination on a CD are

reduced

. Light from a laser is focused onto the data-carrying surface of the disk. The size of the spot

on the pits and lands is 1.7

m

m, whereas the size of the spot on the upper clear surface of the disk is 800

m

m. The spot on the clear surface is nearly 500 times larger than the spot on the pits and land. This means that the system is relatively tolerant of surface contamination because slight imperfections are out of focus.Slide99

© 2014 Cengage Learning Engineering. All Rights Reserved.

99

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Disk Speed The speed of an audio CD is governed by the speed at which data is required from the disc.

This speed corresponds to a data rate of 150 Kbits/s and is called 1X. At this speed, the surface of the disk moves under the read head at 1.25 m/s.

Because computer users want to read data as fast as possible, CD drives have become faster since their introduction. A 4X drive provides data at four times the rate of a standard audio disc. Drives operating at 48X are now commonplace.

However

, the test results published by organizations that benchmark drives demonstrate that these drives don’t provide the sustained data rates you might expect.

A

48X disk is not 48 times faster than a 1X disc

.Slide100

© 2014 Cengage Learning Engineering. All Rights Reserved.

100

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

First-generation CD drives operated at a constant linear

velocity, CLV, unlike the hard disk. A constant liner velocity means that the speed of the surface of the disk under the read head is constant. A hard disk operates at a constant angular

velocity. Since the radius of the tracks at the center and edge of the disk are markedly different, the rotational speed required to read the track at a constant rate varies as the disk is read. This feature limits the speed of the CD ROM drive.Modern CD ROM drives have largely abandoned pure CLV because it is quite difficult to implement at high rotational speeds.

Some

drives use a dual mode in which their angular velocity is constant when reading tracks close to the center, whereas their linear velocity is constant when reading tracks closer to the outer edge. Slide101

© 2014 Cengage Learning Engineering. All Rights Reserved.

101

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The Optical Read-head

 

Figure 11.44 illustrates the path taken by the light from the laser to the surface of the disc and then back to the photoelectric sensors. Light from the laser

is

focused to a spot on the disk.

Light

from the spot is reflected back along the same path; the amount of light reflected depends on whether the spot is hitting a pit or land

.Slide102

© 2014 Cengage Learning Engineering. All Rights Reserved.

102

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

When

the light returns from the disc, it hits a beam splitter and some of it is reflected down to the sensor where it is detected. A photodiode measures the amount of light being reflected from the surface of the disk.

If the light from the surface of the disk is coming from a bump, some of the light has traveled a further

l

/2 and is 180

out of phase resulting in a fall in the signal level at the detector. Slide103

© 2014 Cengage Learning Engineering. All Rights Reserved.

103

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Focusing and Tracking To read data reliably, an optical drive’s read head must follow the spiral track accurately.

It is easy to move the device optics radially on runners to step in or out along a radius.

It is harder to move the beam to the required spot with the necessary precision. This applies to movement in the X-Y plane when seeking a track and in the Z plane when focusing.The objective lens in the read head is mounted on gimbals and can move in two planes; left and right for tracking, and in and out for focusing.

A

magnetic field from an electromagnet is used to position the lens to perform the fine tracking and focusing.Slide104

© 2014 Cengage Learning Engineering. All Rights Reserved.

104

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure

11.45 illustrates

the optical sensors. Light from the laser passes through a diffraction grating, a transparent material indented with parallel lines. The

diffraction

grating

splits

the

beam

into a main beam and two side beams

.

Sensors A, B, C, and D pick up the main spot.

Sensors

E and F pick up the two side beams. The outputs of the two side beam sensors are subtracted to get

track

error

= E – F. If the beam is centered, the tracking error is zero. If

track

error

> 0 the beam must be moved left, and if

track

error

< 0 the beam must be moved right.Slide105

© 2014 Cengage Learning Engineering. All Rights Reserved.

105

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The main beam falls on sensors A, B, C, and D where the sum of the outputs of these sensors is used to regenerate the data from the disk. Differences in outputs between pairs of sensors are used to focus the spot

.

In Figure 11.45 the beams from the diffraction grating pass through a collimating lens that makes the beams parallel, a

quarter wave plate

, and an objective lens that focuses the beams onto the disk. These beams are reflected back along the same path they took, and then through the beam splitter to the six photo sensors

.Slide106

© 2014 Cengage Learning Engineering. All Rights Reserved.

106

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The

objective lens is

cylindrical rather than spherical, which introduces astigmatism in the focusing; that is, the focal point is different in the vertical and horizontal planes. If the beam is in focus, the spot is circular and all four central sensors receive equal amounts of energy. If the objective is too close to the surface, the beam is elliptical, as

Figure

11.45b demonstrates, and the signal (A+D) is greater than (B+C). The difference can be used to move the lens back from the disk. If, however, the lens is too far away from the surface, the effect of the astigmatism is to rotate the elliptical spot by 90

0

and make (B+C) greater than (A+D).Slide107

© 2014 Cengage Learning Engineering. All Rights Reserved.

107

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Low-level Data Encoding The encoding of data on a disc is subject to several severe constraints, the most important of which concerns the

distribution of energy in the spectrum of the signal from the optical pickup. In

particular, there should be no DC component in the signal; that is, its average value must be zero. The low-level data encoding must be designed to control the number of consecutive zero bits or one bits, and allow the clock signal to be regenerated from the data signal. Source data is stored in units of eight-bit bytes.

These

data bytes are each encoded into fourteen bits by means of 8-to-14-bit modulation, EFM. Each 14-bit code can represent 2

14

= 16,376 different values, although only 2

8

= 256 of these values are used.

EFM

is a form of

run length limiting

code that reduces the bandwidth of the signal. Slide108

© 2014 Cengage Learning Engineering. All Rights Reserved.

108

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The EFM ensures that there are no more than ten and no less than two consecutive zeroes in the data stream on the disc.

In fact, there are 267 legal 14-bit patterns conforming to the rule: the number of consecutive zeros in the data stream must be between three and nine, inclusive.

This means that there are 267 – 256 = 19 bit patterns that are legal from the modulation mechanism, but that do not describe a valid data byte. Some of these codes can, therefore, be used as special markers in the data stream.Slide109

© 2014 Cengage Learning Engineering. All Rights Reserved.

109

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Unfortunately, it is possible for the end of one group and the beginning of the next group to be misinterpreted as a valid code word.

To stop this happening, a 3-bit delimiter is placed between each group of 14 bits. These three bits are called

merging bits. The three bits chosen to form the merging pattern are not always the same. The specific merging pattern is selected to ensure that the average number of 1s and 0s remains approximately the same (this constraint ensures that there is no DC component in the data from the optical head).Slide110

© 2014 Cengage Learning Engineering. All Rights Reserved.

110

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

A logical 1 value in the data is interpreted as a change of state in the signal from the disk (i.e., a transition from land to pit or vice versa) so that the 1s are represented by the starts and ends of recorded pits.

Figure

11.46 illustrates the relationship between the data stream and the disk surface.Slide111

© 2014 Cengage Learning Engineering. All Rights Reserved.

111

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.47 illustrates the three levels of structure on an audio CD. In

11.47(a

) the block is composed of 98 frames. Each frame contains a single-byte subcode that contains information about the current track etc.

Figure

11.47(b) shows the frame structure with its 24 bytes of audio data, synchronizing header,

subcode

bytes and the parity check bytes generated by the Reed Solomon encoder.

This

Figure is misleading because the order of the data is scrambled by the CIRC coder.Slide112

© 2014 Cengage Learning Engineering. All Rights Reserved.

112

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Re-writable CDs To make CD-RW technology compatible with existing CD drives, it was necessary to find a means of creating and deleting areas of differing reflectivity along the track of a disc.

Two candidate technologies were phase-change

and magneto-optical. Panasonic and others pioneered magneto-optical storage and such devices are widely used today. Phase-change technology has been universally adopted for CD-RW devices

.

Increasing the power of a tightly focused laser beam locally heats the surface of the data-carrying layer

.

This layer contains a compound of silver, indium, antimony and tellurium that

exists

in two stable states, a crystalline state and an amorphous state.

When

this material is crystalline, it reflects the laser light better than when it is amorphous. Slide113

© 2014 Cengage Learning Engineering. All Rights Reserved.

113

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The CD-RW disc itself is similar to the conventional CD. The

substrate is a 1.2 mm polycarbonate disc and the track (i.e., spiral groove) is molded on the disk with the time information. The recording layer is sandwiched between two dielectric layers

to control the thermal characteristics of the phase-change layer when it is heated during the writing or erasing process. A reflective layer is provided behind the data and dielectric layers.The laser in a CD-RW drive operates at three powers. During reading, it provides the beam that detects the edges of the pits and operates in its lowest power mode.

During

writing, the laser operates in its highest power mode when it heats the recording layer sufficiently to create an amorphous state. The write power level heats the surface locally to about 600

0

C and the compound melts. When it cools rapidly, the liquid freezes and shrinks to create a “pit”.Slide114

© 2014 Cengage Learning Engineering. All Rights Reserved.

114

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

When the CD-RW drive is erasing data, the laser operates in a lower power mode than its write mode and the laser heats the surface sufficiently to turn the data layer into its crystalline state.

The phase-change material is heated to about 2000 C where it crystallizes and the atoms take up an ordered state.

There is a limit to the number of write and erase cycles that the material can undergo and still continue to provide two optically distinguishable states. CD-RW is still a read-mostly medium rather than a true read/write medium.Slide115

© 2014 Cengage Learning Engineering. All Rights Reserved.

115

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Magneto-optical Storage An alternative to the phase-change technology is

magneto-optical recording. Magneto-optical systems are not fully compatible with conventional CD drives and the rise of the low-cost CD-RW drive has led to a decline in MO technology.

Recall that the Curie temperature defines the point at which a magnetic material loses its magnetization. Some substances have a Curie point of 2000 C, which means that they can be demagnetized by heating with a laser. Slide116

© 2014 Cengage Learning Engineering. All Rights Reserved.

116

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Figure 11.48 illustrates the principle of a MO system. The data-carrying surface of the disc is a ferromagnetic material with a low Curie point. In normal operation, the domains on the surface are magnetized perpendicular to the disc’s surface.

If

the surface is heated by a laser, it is demagnetized. However, because there is an electromagnet under the disc, applying a field will magnetize the surface when it cools.Slide117

© 2014 Cengage Learning Engineering. All Rights Reserved.

117

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

An MO disc can be read because there is an interaction between magnetism and optics. When a polarized light passes through a magnetic material, a change in the polarization of the light takes

place (the

Kerr effect).

A similar optical system to that in a CD can be used to read MO discs. When light is reflected from the disk, its plane of polarization is rotated by about 0.5

0

if the surface is magnetized in one direction, and by 0.5

0

in the other direction if the magnetization is reversed. Consequently, there is a difference of 1

0

in the polarization of the light from the two magnetic states. Slide118

© 2014 Cengage Learning Engineering. All Rights Reserved.

118

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The DVD The DVD is a development of the compact disc that means

Digital video disc or Digital versatile disc.

The DVD was designed to store a movie. Unlike the CD-ROM, the DVD is available in several capacities depending on whether there are one or more data-carrying layers. DVD technology was developed in the early 1990s by a group of companies including Toshiba, Time Warner, Sony and Philips

.

Some of the leading players in the development of DVD technology had close links with Hollywood, which strongly influenced the emerging standard. In particular, the DVD was designed to provide 133 minutes of encoded video information (sufficient to cover most mainstream movies).

The

DVD provides

high-quality

sound and audio and includes up to three separate audio channels, allowing the same DVD to be used with audiences of different nationalities. Slide119

© 2014 Cengage Learning Engineering. All Rights Reserved.

119

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

DVD technology is virtually the same as CD technology.

In

fact, you could say that the DVD is the CD constructed with technology that’s advanced and matured over ten years.

The

pits on DVD discs are packed

tightly

and the minimum pit size is 0.4

m

m rather than 0.8

m

m used on a CD. The laser light wavelength is reduced from 780 nm to 640 nm.

The

track spacing is reduced from 1.6

m

m to 0.75

m

m. Figure 11.49 illustrates the structure of CD and DVD tracks. Slide120

© 2014 Cengage Learning Engineering. All Rights Reserved.

120

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

The thickness of the DVD disc is 0.6 mm, or half that of a CD disc.

The

DVD is made double sided by bonding together two 0.6 mm discs to create a double-sided 1.2 mm disc.

Moreover

, it is possible to have two different data layers in the same disc.

DVD

supports the following four basic formats illustrated by Figure 11.50.Slide121

© 2014 Cengage Learning Engineering. All Rights Reserved.

121

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Recordable DVDs Recordable

DVD technology didn’t settle down as rapidly as CDs where the CD, CD-R, and CD-RW followed each other with a reasonable degree of backward compatibility. Within

a few years the DVD reader was followed by the DVD-RAM, DVD-RW, DVD+RW and DVD-R.DVD-R is a write-once medium with a capacity of 4.7 or 9.4 GB that can be used in most compatible DVD drives. It first appeared in 1997 with a lower capacity 3.95 GB version. In 1989 the first rewritable device appeared, the DVD-RAM, that relied on both phase-change and magneto-optical techniques to write data to the disc.

First

generation devices had a 2 GB capacity but that rose to 4.7 GB by 1999.This system was not compatible with other DVD formats. Slide122

© 2014 Cengage Learning Engineering. All Rights Reserved.

122

Blu-ray Just as the DVD replaced the CD, Blu-ray technology is replacing the DVD. Blu-ray was driven by the introduction of high definition television, HDTV, that required more storage capacity than the DVD could provide.

Without a new storage medium, high definition home cinema would have been impossible (other than via off air broadcasting). Alas, two different solutions to the problem of increasing optical media capacity were proposed: HD DVD and Blu-ray.

Both systems were launched, each backed by media giants. Blu-ray was championed by Sony, Panasonic, Philips, LG, Pioneer and Apple. HD DVD was championed by Toshiba, Hitachi, Microsoft, and NEC.

Computer Organization and Architecture: Themes and Variations, 1st Edition ClementsSlide123

© 2014 Cengage Learning Engineering. All Rights Reserved.

123

Two incompatible formats provided an echo of the struggle between the VHS and Betamax standards for VCR tapes decades earlier. Two formats would have forced the consumer to make a choice and stores to stock films in each of the formats. Ultimately

, the major studios had the greatest bargaining power in the battle of standards. Sony Pictures, MGM, Disney, and 20th Century Fox selected Blu-ray and only Universal Studios (with about 9% of the market) chose HD DVD.

Sony also chose Blu-ray for its popular PlayStation 3 gaming console (with 3.2 million consoles in the USA alone) to increase the demand for Blue-ray products. Another nail in the coffin of HD DVD was Wal-Mart’s promotion of Blue-ray.

Blu-ray prevailed and the world was spared a long-lasting format battle. Computer Organization and Architecture: Themes and Variations, 1

st

Edition ClementsSlide124

© 2014 Cengage Learning Engineering. All Rights Reserved.

124

Blu-ray achieves its high storage density of 25 GB (i.e., 5.3 times that of DVD) by using a higher frequency laser with a wavelength of 405 nm. The visible spectrum extends from 620 nm (red) to 450 nm (violet) which means that the Blu-ray laser is blue/violet and hence its name.The

DVD laser is 650 nm (red) and the CD laser at 780 nm falls in the infra-red spectrum. Blue-ray disk are physically different to CDs and DVDs because the Blu-ray data layer is below a front cover layer of only 0.1 mm.

The numerical aperture of lenses used to focus the beam in Blu-ray systems is higher (0.85) than for DVD (0.6) and CD (0.45). Recall that a high numerical aperture allows a smaller spot size.

Computer Organization and Architecture: Themes and Variations, 1

st

Edition ClementsSlide125

© 2014 Cengage Learning Engineering. All Rights Reserved.

125

Computer Organization and Architecture: Themes and Variations, 1st Edition Clements

Like DVD, Blu-ray supports dual-layers to provide a capacity of 50 GB.

Figure

11.51 provides images of the three generations of recording media demonstrating their relative densities.